Viscosity Modeling and Prediction of Amorphophallus oncophyllus and Sapindus rarak Using Machine Learning Methods - Study

Aug 3, 2026

Upload by Eden

Viscosity governs fluid mobility in reservoir chemical flooding, directly influencing oil sweep efficiency and overall oil recovery. This study applies machine learning models—ANN and ANFIS—to build correlations for viscosity of polymer solutions prepared from Amorphophallus oncophyllus and Sapindus rarak. Viscosity data from 21 samples with varying concentration and salinity are digitized for modeling. Both methods accurately predict viscosity, achieving average correlation coefficients of 0.997240 (ANN) and 0.995124 (ANFIS).

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Introduction

Viscosity role in oil reservoir mobility and chemical flooding

Natural polymer and surfactant sources (porang and lerak)

Motivation for machine learning viscosity correlations

Methods and Modeling

ANN and ANFIS for viscosity prediction

Data preparation and augmentation

Results and Performance

Prediction accuracy using correlation coefficients

Conclusion

Effectiveness of ANN and ANFIS for viscosity estimation

JFA (JURNAL FISIKA DAN APLIKASINYA) VOLUME 21, NUMBER 1, JANUARY 2025 Viscosity Modeling and Prediction of Amorphophallus oncophyllus and Sapindus rarakusing Machine Learning Methods Muhammad Tauﬁq Fathaddin*, Dwi Atty Mardiana, Andrian Sutiadi, and Fajri Maulida Department of Petroleum Engineering, Universitas Trisakti, Jl. Kyai Tapa No. 1, Jakarta, 11440 Abstract: Viscosity plays an important role in regulating the mobility of ﬂuids injected into the reservoir to increase the efﬁciency of oil sweeping. This study discusses the application of Machine Learning methods, namely ANN and ANFIS, to model the correlation of physical properties of Amorphophallus oncophyllus and Sapindus rarak solutions. The purpose of this study is to obtain a correlation to determine the viscosity of the polymer solutions. The data used include viscosity measurements for 21 samples of Amorphophallus oncophyllus and Sapindus rarak solutions with variations in concentration and salinity. The data is augmented by digitization for modeling. The results show that both Machine Learning methods can estimate viscosity values well. Very accurate results are achieved by applying ANN and ANFIS with average correlation coefﬁcients of 0.997240 and 0.995124, respectively. Keywords: Chemical ﬂooding; concentration; oil recovery; salinity; viscosity *Corresponding author: muh.tauﬁ[q@trisakti.ac.id](q@trisakti.ac.id) [http://dx.doi.org/10.12962/j24604682.v21i1.21953](http://dx.doi.org/10.12962/j24604682.v21i1.21953) [2460-4682](2460-4682) 􀀍cDepartemen Fisika, FSAD-ITS I. INTRODUCTION Viscosity plays an important role in regulating the mobility of ﬂuids injected into oil reservoirs. To maximize oil sweep, the mobility of the injected ﬂuid should be lower than that of the displaced ﬂuid. The mobility of the injected ﬂuid that is higher than that of the displaced ﬂuid (oil) will cause a ﬁngering phenomenon, where the injected ﬂuid penetrates into the displaced oil zone. This phenomenon causes the injected ﬂuid to bypass the oil zone, so that the oil sweep is not optimal, which causes a low oil recovery factor. Chemical substances such as polymers and surfactants added to the injected ﬂuid will affect its viscosity. The amount of substances added to the injected ﬂuid must be appropriate. These substances are needed to increase the viscosity of the injected ﬂuid so that the mobility of the injected ﬂuid is lower than that of the displaced ﬂuid, but the addition of the viscosity of the injected ﬂuid should not be too high, which causes the mobility to be too low. The viscosity of the injected ﬂuid that is too high also causes a low oil recovery factor. High viscosity can make it difﬁcult to ﬂow ﬂuids. Therefore, to ﬂow ﬂuids requires a higher injection pressure. High injection pressure risks damaging reservoir rocks and injection equipment. In addition, mobility that is too low risks causing the deposition of these chemicals in the pores of the rock. This results in a decrease in the concentration of substances in the solution and results in a decrease in rock permeability [1-6] . Chemical substances such as polymers and surfactants can be obtained from natural or artiﬁcial materials. In this study, porang and lerak were used as polymers and surfactants. Porang (Amorphophallus oncophyllus) is one of the indigenous Indonesian Amorphophallus plants whose tubers are very potential as a source of glucomannan. Yellow porang tubers (Amorphophallus oncophyllus Pr) contain around 55% glucomannan, while white porang tubers (Amorphophallus variabilis Bl) contain around 44% glucomannan [7-8] . Glucomannan is used as a thickener, gel former, texture improver, water binder, stabilizer and emulsiﬁer [9] . Glucomannan from porang tubers has been cultivated in Indonesia, but the number of studies is still relatively small on its properties and potential applications [7] . Lerak (Sapindus rarak) are plants that are used as raw materials for natural soap because

FAQs

Why is viscosity important in chemical flooding for oil recovery?

Viscosity controls the mobility of injected fluid compared with displaced oil. If injected mobility is too high, it causes fingering that reduces sweep efficiency and lowers oil recovery; if too low, it can cause excessive chemical deposition and reduced permeability.

What machine learning methods are used to predict viscosity in this study?

The study uses two machine learning approaches: ANN (Artificial Neural Network) and ANFIS (Adaptive Neuro-Fuzzy Inference System) to model the viscosity of the polymer solutions.

How is the dataset prepared for modeling viscosity?

Viscosity measurements from 21 samples of Amorphophallus oncophyllus and Sapindus rarak solutions are collected with variations in concentration and salinity. The data is digitized and used to train and evaluate the ANN and ANFIS models.

Aug 3, 2026

Upload by Eden